Opportunities and limitations of the GPT store in Financial Services

OpenAI has recently announced the launch of the GPT Store, giving creators a way to easily share and monetise their custom GPTs.  

Whilst Open AI still hasn’t clarified how exactly GPTs will be monetised, this initiative opens up exciting opportunities, especially for those skilled in crafting GPTs for niche uses. But what does it mean for organisations in the financial services sector looking to increase adoption and productivity with Generative AI?

In this article, we take a deep dive in the opportunities and pitfalls of GPTs for enterprise usage within the financial services industry.  

What is a GPT?

Think of a ‘GPT’ as a pre-generated bundle of knowledge and prompts created to do a specific task. If the task is generic enough, and ‘solvable’ through clever prompting and some knowledge, a GPT is essentially a transferrable solution that can be re-used within an enterprise from one team to the next. It is especially useful for tasks that are often repeated by different teams, perhaps in different contexts. Think of it as a shared code library for example, that is useful to assess or price financial instruments across different teams. However rather than it being a numeric tool, think of it as a linguistic activity that is shared.

For example, a research team within a bank or asset management firm will need to assess market data and information from a varied set of sources to understand how it relates to their investment thesis. These market views and sentiments are expressed in language, and loosely follow a kind of ‘forecast thesis and evidence’ whereby summaries of a thesis are put forward, and evidence is shown to support that thesis. This is done for a varied set of asset classes from equity, fixed income, private equity, infrastructure, to more complex derivative products. In each case, it can be done differently. The example below showcases how this may be integrated into a GPT itself.

Opportunities of the GPT Store for Financial Services

Ease of Access and Variety:

• Employees can easily access a diverse range of AI models, tailored to specific use cases. Rather than having to learn the ins and outs of prompting and starting from 0, they can find GPTs already trained to assist with a wide range of tasks including customer communications, competitor analysis and market updates.

Rapid Deployment and Innovation:

• The store enables quick deployment and scaling of GPT models, crucial for adapting to dynamic market conditions in the financial sector. Organisations will be able to leverage the innovation of GPT creators from other organisations.

Enhancing Employee Creativity:

• Use-case-centric GPTs can inspire employees to explore AI solutions, fostering innovative problem-solving in financial operations. Being exposed to clever GPTs will encourage employees to look beyond using Chat GPT for basic tasks such as writing emails.

Consistent Customer Interactions:

• Pre-defined GPTs ensure consistent communication standards in customer-facing operations, vital for maintaining professional service levels. They can even be instructed to provide specific options to start a conversation. But if you want the GPT to draw on knowledge specific to your organisation, or to use a tone of voice consistent with your brand guidelines, you’d have to create your own custom GPT.

Limitations and Risks of the GPT Store for Enterprises

Non-Customisability and Limited Tailoring:

• The store's models may not be sufficient for specific, complex financial tasks that require bespoke solutions such as being trained on a specific knowledge base about the different products you offer.

Transparency and Replicability Concerns:

• The opaque nature of GPT model operations and replicability issues can pose challenges for financial institutions needing clear audit trails.

Security and Compliance Uncertainties:

• Questions remain about how well these models align with the specific security and compliance requirements of the financial sector. If you’re relying on use cases that involve sensitive customer data, it might make sense to explore a more bespoke solution with a dedicated software engineering team.

Quality and Relevance Challenges:

• Identifying high-quality, relevant GPTs that match specific use cases amidst a plethora of options is a significant challenge. Creating custom GPTs that are much more tailored to your organisation will be a great skill for your workforce to possess in 2024.

Looking forward

Where 2023 was about the discovery of the capability of generative AI, 2024 will be about finding numerous tasks that can be augmented or even replaced by AI.

The GPT Store offers promising opportunities for the financial services sector, from easy access to AI models to enhancing employee creativity and inspiring more tailored use cases. But if you’re looking for scalable enterprise solutions for specific use cases unique to your organisation, you might be better placed to partner with a team with expertise in developing custom use cases that will drive productivity across your organisation.